Click on “Download PDF” for the PDF version or on the title for the HTML version.

If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options.

Improvement of field phenotyping from synchronized multi-camera image collection based on multiple UAVs collaborative operation systems

Published by the American Society of Agricultural and Biological Engineers, St. Joseph, Michigan

Citation:  2022 ASABE Annual International Meeting  2200268.(doi:10.13031/aim.202200268)
Authors:   Hyeon-Seung Lee, Mr., J. Alex Thomasson, Prof., Xiongzhe Han, Prof.
Keywords:   UAV, Remote sensing, Multiple UAV, Field phenotyping

Abstract. The technology using multiple unmanned aerial vehicles has been intensively applied in the national defense and the festival (event) but has not been carried out to improve the agricultural field. To successfully implement precision agriculture for efficient crop production, three-dimensional (3D) reconstruction of the plant canopy is essential to measure geometry such as height, width, volume, and leaf coverage. However, image data collection based on a single unmanned aerial vehicle is challenging to acquire an accurate crop shape depending on the angle of sunlight, and it is difficult to accurately measure leaves and stems in an environment in which crops are grown closely. To improve the 3D models to minimize errors caused by wind and lack of detailed information in overlapping regions, an image-based 3D plant reconstruction system based on multiple unmanned aerial vehicles is proposed to achieve two images simultaneously from different views of the plants during flight. An unresolvable offset may produce a stereo view of the environment if unsynchronized cameras capture nonstationary objects. Therefore, camera synchronization is required for the simultaneous acquisition of images from different perspectives to compensate for errors in plant-height measurement. This study will reconstruct 3D models of the crop with Structure-from-Motion based on a multiple-view stereo algorithm and metric construction and compare errors between traditional 3D models from single-camera image collection with errors from the synchronized multi-camera image collection.

(Download PDF)    (Export to EndNotes)